Due to the development of artificial intelligence (AI) technology, the problem of singularity has taken an important place in new studies on economic growth. Many expect that in the coming decades, AI will be able to reach the level of human intelligence and take over all or most of the tasks performed by people today. This will cause an economic growth explosion so that economy could transit into a singularity regime with superexponential growth rates (in the limit, approaching to infinity). The paper examines in detail the arguments both for and against such an over-optimistic scenario. The arguments in its favor are based on the key assumption of endogenous growth models that the ultimate source of economic growth is ideas (scientific and technological knowledge). The basic theoretical counterargument refers to the so-called “Baumol cost disease”, whereby over time the contribution to GDP of sectors with the fastest productivity dynamics steadily decreases. The main empirical counterargument is that if the singularity were already near, the first signs of it would be visible in an unusual behavior of some key macroeconomic parameters. This implies that even if explosive growth ever become a reality, it would happen not soon.